Yu et al. Critical Care 2014, 18:531 http://ccforum.com/content/18/5/531
RESEARCH
Open Access
Propensity-matched analysis of the effect of preoperative intraaortic balloon pump in coronary artery bypass grafting after recent acute myocardial infarction on postoperative outcomes Pey-Jen Yu1*, Hugh A Cassiere1, Sophia L Dellis1, Nina Kohn2, Frank Manetta1 and Alan R Hartman1
Abstract Introduction: There is substantial variability in the preoperative use of intraaortic balloon pumps (IABPs) in patients undergoing coronary artery bypass grafting post myocardial infarction. The objective of this study is to determine the effect of preoperative IABPs on postsurgical outcomes in this subset of patients. Methods: From 2007 to 2012, 877 patients underwent isolated coronary artery bypass post myocardial infarction. Four hundred and six patients were propensity-score matched based on the likelihood of receiving a preoperative balloon pump. Total blood transfusion requirements, composite in-hospital morbidity and/or mortality end point, total hours in the intensive care unit, and length of hospital stay were compared between the two groups. Results: No significant differences in demographics, preoperative risk factors, intraoperative variables or length of hospital stay were found between patients with and without balloon pumps after propensity score matching. Compared to patients without balloon pumps, a higher percentage of patients with preoperative IABPs required transfusions. Patients with preoperative balloon pumps were more likely to have the composite end point of in-hospital morbidity (24.1% versus 12.8%, P 2.0 liters/min/m2 unless there is hypotension or systemic manifestations of a low output state. Sepsis is defined by the presence of a Systemic Inflammatory Response Syndrome resulting from suspected or proven infection. Preoperative anticoagulation is defined as the use of intravenous or subcutaneous thrombin inhibitors within 48 hours of surgery. Gastrointestinal complication is defined as gastrointestinal bleeding, perforation, or infarction. Postoperative respiratory insufficiency is defined as failure to wean from the ventilator within 24 hours postoperatively. STS surgical risk scores were calculated using the online risk calculator [1]. To determine the baseline perioperative risk independent of the presence of an IABP for each individual patient, a modified STS risk score was calculated in which the presence of an IABP was not factored into the calculation. Out of the 877 patients, 346 received an IABP preoperatively and 531 did not receive an IABP preoperatively. Five patients (two with an IABP and three without an IABP) had missing data and were excluded from further analysis. Of the 872 patients with complete data, 406 were propensity-score matched based on the likelihood of receiving an IABP preoperatively (203 with an IABP, 203 without an IABP). Clinical end points included total blood transfusion requirements, composite inhospital morbidity and/or mortality, total hours in the intensive care unit, and length of hospital stay. Composite in-hospital morbidity consisted of stroke, deep sternal wound infection, respiratory failure, reoperation for bleeding, new postoperative renal failure, sepsis, vascular complications including dissection and limb ischemia, and postoperative myocardial infarction. All statistical analyses were performed using SAS Version 9.3 (SAS Institute Inc., Cary, NC, USA). Data analysis was performed retrospectively. Propensity score matching was used to match preoperative IABP patients to controls (non-preoperative IABP patients) on several potentially confounding variables. Propensity score matching is an efficient alternative to matching on individual variables [2]. The probability of receiving a preoperative IABP (that is the propensity score) was calculated using a logistic regression model. Factors included in the model were: age, gender, body mass index, left ventricular ejection fraction (LVEF), preoperative creatinine, presence of comorbidities (cerebral vascular disease, diabetes mellitus, hypertension, smoking, hyperlipidemia, peripheral vascular disease, dialysis, congestive heart failure, chronic obstructive pulmonary disease), preoperative use of
Yu et al. Critical Care 2014, 18:531 http://ccforum.com/content/18/5/531
antiplatelet agents and/or anticoagulants including intravenous heparin, use of intravenous nitroglycerine, reoperation, time of between myocardial infarction and CABG, coronary artery disease burden (number of vessels with >70% stenosis, left main coronary artery disease >50%), concurrent valvular disease, and preoperative hemodynamic instability. Each patient was matched to a single control, based on the propensity score, using the SAS macro OneToManyMTCH [3]. Conditional logistic regression was then used to compare binary outcomes between the resulting matched pairs. The Wilcoxon signed-rank test was used to compare continuous outcomes between the matched pairs. The Mann-Whitney test was used to evaluate the comparability of continuous factors used to generate the propensity score between the two groups both prior to and after matching. Similarly, the chi-square test was used to evaluate the comparability of categorical factors.
Results Preoperative characteristics for the 877 patients undergoing CABG after an acute myocardial infarction are listed in Table 1. A total of 39.5% of patients received a preoperative IABP. Those receiving a preoperative IABP were younger with fewer comorbidities. In particular, patients with a preoperative IABP had less preoperative renal insufficiency, hypertension, cerebrovascular disease, and peripheral vascular disease. However, preoperative IABPs were placed more often in patients with hemodynamic instability, lower ejection fractions, left main coronary artery disease, and in patients on intravenous nitroglycerin. Patients with IABPs were more likely to undergo CABG on an emergent basis and within 24 hours of their myocardial infarction. No significant differences in demographics or preoperative risk factors were found between patients with and without an IABP after propensity score matching (Table 2). Patients with and without an IABP had similar rates of prior percutaneous coronary interventions (18.2% vs. 16.3%, P = 0.60) after propensity matching. Similarly, there was no statistically significant differences in the rate of percutaneous coronary interventions performed immediately before CABG (7.4% vs. 3.4%, P = 0.08). There was no difference in the use of antiplatelet and anticoagulation agents between the two cohorts after propensity matching (Table 2). In particular, there was no difference in the preoperative use of GPIIa/IIIb inhibitors (2.9% vs. 2.5%, P = 0.76). The modified STS predicted risks of mortality and combined morbidity and mortality were also comparable after propensity score matching (Table 2). Although the groups were matched only on preoperative characteristics, there were no major differences in the intraoperative characteristics between the two groups after matching (Table 3). Specifically, there were no statistically
Page 3 of 8
Table 1 Pre-matched characteristics of patients No preoperative Preoperative P value IABP (n = 531) IABP (n = 346) Age (years)
67.0 (60.0, 75.0)
64.5 (57.0, 74.0) 0.008
Female
145 (27.3%)
83 (24.0%)
Body mass index
27.6 (24.8, 31.2)
26.8 (24.4, 29.8) 0.076
0.274
Cerebrovascular disease
110 (20.7%)
54 (15.6%)
0.058
Peripheral vascular disease
73 (13.8%)
33 (9.5%)
0.062
Diabetes
273 (51.4%)
164 (47.4%)
0.245
Congestive heart failure
157 (29.6%)
109 (31.5%)
0.542
Dialysis
28 (5.3%)
9 (2.6%)
0.054
Preoperative creatinine
1.10 (0.90, 1.50)
1.05 (0.90, 1.30) 0.021
Chronic obstructive lung disease
163 (30.7%)
120 (34.7%)
0.217
Hypercholesterolemia
228 (42.9%)
155 (44.8%)
0.587
Hypertension
442 (83.2%)
241 (69.7%)
RESEARCH
Open Access
Propensity-matched analysis of the effect of preoperative intraaortic balloon pump in coronary artery bypass grafting after recent acute myocardial infarction on postoperative outcomes Pey-Jen Yu1*, Hugh A Cassiere1, Sophia L Dellis1, Nina Kohn2, Frank Manetta1 and Alan R Hartman1
Abstract Introduction: There is substantial variability in the preoperative use of intraaortic balloon pumps (IABPs) in patients undergoing coronary artery bypass grafting post myocardial infarction. The objective of this study is to determine the effect of preoperative IABPs on postsurgical outcomes in this subset of patients. Methods: From 2007 to 2012, 877 patients underwent isolated coronary artery bypass post myocardial infarction. Four hundred and six patients were propensity-score matched based on the likelihood of receiving a preoperative balloon pump. Total blood transfusion requirements, composite in-hospital morbidity and/or mortality end point, total hours in the intensive care unit, and length of hospital stay were compared between the two groups. Results: No significant differences in demographics, preoperative risk factors, intraoperative variables or length of hospital stay were found between patients with and without balloon pumps after propensity score matching. Compared to patients without balloon pumps, a higher percentage of patients with preoperative IABPs required transfusions. Patients with preoperative balloon pumps were more likely to have the composite end point of in-hospital morbidity (24.1% versus 12.8%, P 2.0 liters/min/m2 unless there is hypotension or systemic manifestations of a low output state. Sepsis is defined by the presence of a Systemic Inflammatory Response Syndrome resulting from suspected or proven infection. Preoperative anticoagulation is defined as the use of intravenous or subcutaneous thrombin inhibitors within 48 hours of surgery. Gastrointestinal complication is defined as gastrointestinal bleeding, perforation, or infarction. Postoperative respiratory insufficiency is defined as failure to wean from the ventilator within 24 hours postoperatively. STS surgical risk scores were calculated using the online risk calculator [1]. To determine the baseline perioperative risk independent of the presence of an IABP for each individual patient, a modified STS risk score was calculated in which the presence of an IABP was not factored into the calculation. Out of the 877 patients, 346 received an IABP preoperatively and 531 did not receive an IABP preoperatively. Five patients (two with an IABP and three without an IABP) had missing data and were excluded from further analysis. Of the 872 patients with complete data, 406 were propensity-score matched based on the likelihood of receiving an IABP preoperatively (203 with an IABP, 203 without an IABP). Clinical end points included total blood transfusion requirements, composite inhospital morbidity and/or mortality, total hours in the intensive care unit, and length of hospital stay. Composite in-hospital morbidity consisted of stroke, deep sternal wound infection, respiratory failure, reoperation for bleeding, new postoperative renal failure, sepsis, vascular complications including dissection and limb ischemia, and postoperative myocardial infarction. All statistical analyses were performed using SAS Version 9.3 (SAS Institute Inc., Cary, NC, USA). Data analysis was performed retrospectively. Propensity score matching was used to match preoperative IABP patients to controls (non-preoperative IABP patients) on several potentially confounding variables. Propensity score matching is an efficient alternative to matching on individual variables [2]. The probability of receiving a preoperative IABP (that is the propensity score) was calculated using a logistic regression model. Factors included in the model were: age, gender, body mass index, left ventricular ejection fraction (LVEF), preoperative creatinine, presence of comorbidities (cerebral vascular disease, diabetes mellitus, hypertension, smoking, hyperlipidemia, peripheral vascular disease, dialysis, congestive heart failure, chronic obstructive pulmonary disease), preoperative use of
Yu et al. Critical Care 2014, 18:531 http://ccforum.com/content/18/5/531
antiplatelet agents and/or anticoagulants including intravenous heparin, use of intravenous nitroglycerine, reoperation, time of between myocardial infarction and CABG, coronary artery disease burden (number of vessels with >70% stenosis, left main coronary artery disease >50%), concurrent valvular disease, and preoperative hemodynamic instability. Each patient was matched to a single control, based on the propensity score, using the SAS macro OneToManyMTCH [3]. Conditional logistic regression was then used to compare binary outcomes between the resulting matched pairs. The Wilcoxon signed-rank test was used to compare continuous outcomes between the matched pairs. The Mann-Whitney test was used to evaluate the comparability of continuous factors used to generate the propensity score between the two groups both prior to and after matching. Similarly, the chi-square test was used to evaluate the comparability of categorical factors.
Results Preoperative characteristics for the 877 patients undergoing CABG after an acute myocardial infarction are listed in Table 1. A total of 39.5% of patients received a preoperative IABP. Those receiving a preoperative IABP were younger with fewer comorbidities. In particular, patients with a preoperative IABP had less preoperative renal insufficiency, hypertension, cerebrovascular disease, and peripheral vascular disease. However, preoperative IABPs were placed more often in patients with hemodynamic instability, lower ejection fractions, left main coronary artery disease, and in patients on intravenous nitroglycerin. Patients with IABPs were more likely to undergo CABG on an emergent basis and within 24 hours of their myocardial infarction. No significant differences in demographics or preoperative risk factors were found between patients with and without an IABP after propensity score matching (Table 2). Patients with and without an IABP had similar rates of prior percutaneous coronary interventions (18.2% vs. 16.3%, P = 0.60) after propensity matching. Similarly, there was no statistically significant differences in the rate of percutaneous coronary interventions performed immediately before CABG (7.4% vs. 3.4%, P = 0.08). There was no difference in the use of antiplatelet and anticoagulation agents between the two cohorts after propensity matching (Table 2). In particular, there was no difference in the preoperative use of GPIIa/IIIb inhibitors (2.9% vs. 2.5%, P = 0.76). The modified STS predicted risks of mortality and combined morbidity and mortality were also comparable after propensity score matching (Table 2). Although the groups were matched only on preoperative characteristics, there were no major differences in the intraoperative characteristics between the two groups after matching (Table 3). Specifically, there were no statistically
Page 3 of 8
Table 1 Pre-matched characteristics of patients No preoperative Preoperative P value IABP (n = 531) IABP (n = 346) Age (years)
67.0 (60.0, 75.0)
64.5 (57.0, 74.0) 0.008
Female
145 (27.3%)
83 (24.0%)
Body mass index
27.6 (24.8, 31.2)
26.8 (24.4, 29.8) 0.076
0.274
Cerebrovascular disease
110 (20.7%)
54 (15.6%)
0.058
Peripheral vascular disease
73 (13.8%)
33 (9.5%)
0.062
Diabetes
273 (51.4%)
164 (47.4%)
0.245
Congestive heart failure
157 (29.6%)
109 (31.5%)
0.542
Dialysis
28 (5.3%)
9 (2.6%)
0.054
Preoperative creatinine
1.10 (0.90, 1.50)
1.05 (0.90, 1.30) 0.021
Chronic obstructive lung disease
163 (30.7%)
120 (34.7%)
0.217
Hypercholesterolemia
228 (42.9%)
155 (44.8%)
0.587
Hypertension
442 (83.2%)
241 (69.7%)
